Web Research
Web Research — What the Internet Knows About SanDisk
The Bottom Line from the Web
External, web-derived intelligence — industry-tracker share data, peer IR filings, and trade-press commentary — confirms that SanDisk is the smallest of the top-five NAND makers and is fighting to scale enterprise SSDs into a market the SK Group (SK hynix + Solidigm) is taking over in real time (Q4 calendar 2025: SK Group 30.2% share growing 75% QoQ vs. SanDisk 4.4% growing 63.6% QoQ, per TrendForce). At the same time, the most important industry-level number — Kioxia's FY26 results released May 15, 2026 (JPY 2.34T revenue, 37% operating margin) — validates that the entire Flash Ventures wafer pool is earning peak-cycle returns and that the JV is now planning a 41% YoY capex jump to roughly $4.5B combined. The single biggest open question the web has not answered is whether SanDisk's five signed New Business Model (NBM) contracts (~$42B remaining performance obligations) contain hard price floors that would survive the next downturn — every specialist who looked at the NBM agreements flagged this as the highest-priority unresolved disclosure.
Provider notice: The dedicated web-research backend (Parallel Task API) was unavailable for this run due to insufficient credit at the account level. All six pre-planned research phases (industry, warren, quant, sherlock, historian, forensic) and the specialist-query follow-up phase returned zero pages. The findings below are sourced from (a) primary peer IR filings that the Competition agent fetched directly (Kioxia, SK hynix), (b) industry-tracker commentary already in the staged peer evidence (TrendForce, TechInsights, Tom's Hardware), and (c) SEC EDGAR-derived insider/board data. Findings that should have been cross-checked against external news flow but could not be are flagged.
What Matters Most
The findings below are ranked by how much each would move an investor's view of SanDisk today.
1. SK Group (SK hynix + Solidigm) is the share-taking competitor — and the public web data is unambiguous
SK Group's combined enterprise-SSD revenue in Q4 calendar 2025 reached $3.26B (30.2% market share), growing 75% quarter-over-quarter — the fastest growth rate among the top five and the closest challenger to Samsung. SanDisk's $440M Q4 2025 enterprise-SSD revenue is 13.5% of SK Group's number, even though SanDisk's own sequential growth (63.6% QoQ) is impressive in isolation. The point is direction: Solidigm has a multi-year qualification head start at hyperscalers, runs on transitioning Intel + SK hynix nodes, and is being funded by SK hynix's HBM cash flow. Source: TrendForce Q4 2025 enterprise SSD tracker, via competition agent's primary-source fetches.
The structural risk: SanDisk's bull case rests on datacenter being the durable margin lever, but the share leader in that exact category is doubling down with HBM-funded capex SanDisk cannot match.
2. Kioxia just disclosed peak-cycle results — and the JV capex is jumping 41%
Kioxia (TSE 285A) filed FY2026 results on May 15, 2026: revenue JPY 2,337,628M, operating profit JPY 870,369M (37.2% margin), net income JPY 554,496M. The same disclosure window indicates Flash Ventures combined capex (SanDisk + Kioxia) is planning to rise ~41% YoY to roughly $4.5B, per TrendForce/Yole-style commentary cited in the staged competition evidence. Two implications: the up-leg is being confirmed across the whole JV (not a SanDisk-only optical effect), but the supply discipline that has anchored 2026 ASPs is starting to bend as both partners spend into capacity. Sources: Kioxia FY2025 results PDF (ssl4.eir-parts.net/doc/285A/tdnet/2815628/00.pdf); TrendForce industry-tracker commentary, June 2026.
3. SK hynix's January 2026 results show NAND is a small slice of a DRAM/HBM story
SK hynix (KRX 000660) announced FY2025 calendar results on January 28, 2026: revenue KRW 97.1T, operating profit KRW 47.2T — a 49% operating margin. The vast majority of that margin came from HBM, not NAND. The reader-level fact: SanDisk has no HBM, no DRAM, and reports a comparable 40.7% operating margin on a pure-NAND book — which means SNDK is over-earning the NAND cycle right now relative to what a pure-NAND business should structurally deliver. When NAND turns, SanDisk has nowhere to hide; SK hynix and Micron will. Source: SK Hynix Newsroom FY25 results press release (news.skhynix.com/sk-hynix-announces-fy25-financial-results/).
4. Insider activity ran 14 Form 4 filings in November 2025 alone — the heaviest single month
EDGAR submissions for CIK 0002023554 show 91 Form 4 filings between February 14, 2025 and June 5, 2026, with the heaviest months being November 2025 (14 filings), May 2026 (11), and March 2025 (12). The November 2025 spike coincides with the annual meeting and Lead Independent Director transition; the May 2026 cluster sits on top of the parabolic share-price move. Detailed transaction codes (P/S/A/F/M) are not in the staged summary — the next-level question Sherlock flagged (whether CEO Goeckeler is running material 10b5-1 sales after the 40x run-up) cannot be answered from the staged data alone and would require direct Form-4-by-Form-4 fetch from EDGAR. Source: SEC EDGAR submissions API for CIK 0002023554.
Forensic concluded "buying back" signal and 0.21% insider ownership; Sherlock graded alignment a B and flagged the combined CEO/Chair role mid-transition as the top concern. Both grades are based on filings; the open external read is who replaced Matthew Massengill as Lead Independent Director and the November 2025 say-on-pay vote percentage — neither was resolvable from external sources during this run.
5. The High Bandwidth Flash (HBF) story has gone quiet
HBF was announced in August 2025 as "a new paradigm for AI inference" — a SanDisk + SK hynix joint development. Earnings transcripts confirm that the first samples are targeted for 2H calendar 2026 and first AI inference devices for early 2027 (per the Investor Conference May 2026 transcript). But HBF has dropped out of earnings prepared remarks since Q4 FY2025. External web confirmation of whether HBF is progressing, delayed, or being repositioned as an SK hynix-led product was not retrievable in this run. If HBF is alive and on schedule, it's the response to SanDisk's "no DRAM, no HBM" gap. If it's quietly slipping, the AI-inference moat thesis weakens. Open question — both Historian and Moat flagged this as high-priority.
6. Kioxia–SanDisk merger speculation continues — but no formal disclosure exists
Trade-press commentary (TrendForce, June 2026, and various industry tracker pieces) periodically floats a SanDisk–Kioxia full combination, often linked to Bain Capital's residual Kioxia stake and a rumored US/Japan joint NAND fab. No public disclosure from either company confirms active merger discussions. The structural logic exists (one wafer pool, two listed equities is awkward), but the timing and terms are speculative. Investors should treat this as a tail catalyst, not a base case. Sources: TrendForce industry commentary; specialist queries flagged this as high-priority but unresolved.
7. The Flash Ventures JV was extended to 2034 — confirmed in a January 2026 announcement
Forensic flagged a January 2026 announcement confirming Flash Ventures terms through 2034, including a $1.165B manufacturing-services payment schedule. This locks in SanDisk's capex-light wafer supply for a full additional cycle. It also locks in the $11.9B of total off-balance-sheet commitments Forensic flagged as a structural distortion of the on-balance-sheet picture (equity $9.2B; off-balance commitments $11.9B). Source: SanDisk press release, January 2026, referenced in forensic-claude analysis.
8. No short interest, no FINRA position data — the squeeze hypothesis can't be tested from the public web
FINRA returned zero rows for SanDisk reported short interest in this staging window. The 20-day ADV is 11.2M shares. The technical-agent question about whether the November 26, 2025 high-volume rejection candle and the +37% one-month / +27% one-week May–June 2026 move are short-covering-driven cannot be answered with the public short-interest tape; supplementary sources (S3 Partners, Ortex) were not staged. Source: data/short_interest/ (FINRA), data/tech/liquidity.json.
9. The peer-set web search confirms what the filings say: SanDisk's structural weakness is no fab ownership
TechInsights 3D NAND layer comparisons and Tom's Hardware HDD/SSD cost ratios — both cited in the Competition agent's primary-source web work — confirm that Samsung (286L V9, 400+L V10 announced), SK hynix (321L), and Micron (276L) are running ahead of the SanDisk/Kioxia BiCS9 at 218 layers. SanDisk's defense is bit-density-per-layer, not absolute layer count. The bull view: density and watts-per-gigabyte matter more than layer optics in hyperscaler RFPs. The bear view: layer count is a header spec, and "218" requires explanation in every customer conversation. Sources: TechInsights 3D NAND layer comparison; Hardwareluxx/ComputerBase architecture commentary; Tom's Hardware HDD/SSD cost-per-TB (16x ratio as of Q1 2026).
10. WDC's 5.1% residual stake — the two-year tax-free restriction expires February 2027
This is a forward-looking overhang Sherlock surfaced and the external web should have confirmed. Western Digital's residual ownership of SanDisk (originally 19.9% at separation, currently 5.1%) is subject to a two-year tax-free distribution restriction that expires February 21, 2027. Any signaled WDC sell-down after that would be a price-relevant overhang. No public 144 filings, lock-up amendments, or signaled monetization plans were retrievable in this run; this remains an open external watch item. Source: SNDK / WDC separation filings, referenced by Sherlock.
Recent News Timeline
Timeline gaps: TrendForce/Yole tracker dates are approximate (typically published within a month of the period end). The November 2025 volume spike date is confirmed from the price tape but the underlying catalyst is not externally cited in the staged data — the technicals query flagging this as "high priority" remains unresolved by external sources.
What the Specialists Asked
Below is the unresolved-question inventory from each specialist's targeted web-search list. Where the question has been partially answered by primary-source filings or peer IR fetches, the answer is summarized. Where the external web answer is missing (because the Parallel API failed), the gap is noted as "External lookup not completed."
Governance and People Signals
External web confirmation of governance signals was limited by the Parallel API failure. The strongest signals come from SEC EDGAR submissions and the internal proxy/governance staging.
Pattern read: 91 Form 4 filings in 16 months works out to roughly 6 per month average, with three pronounced spikes — March 2025 (post-spin initial grants), November 2025 (annual meeting / LID transition), and May 2026 (Q3 FY26 results + parabolic move). The November and May clusters are the two windows that warrant deeper Form-4-by-Form-4 fetching to determine whether NEOs (Goeckeler, Visoso) are executing 10b5-1 sales of material size — Sherlock flagged this as a critical unresolved external read, and the staged summary does not include the transaction-code detail. Source: SEC EDGAR submissions API for CIK 0002023554.
Open governance items the external web did not confirm in this run:
Who replaced Matthew Massengill as Lead Independent Director after the November 2025 annual meeting.
Say-on-pay vote percentage at the November 2025 meeting (>90% would tip Sherlock's grade to A-minus).
10b5-1 plan adoption and execution by CEO Goeckeler and CFO Visoso in calendar 2026.
Whether Western Digital has signaled timing of its 5.1% residual stake monetization ahead of the February 2027 tax-free restriction expiry.
Compensation Discussion and Analysis details around the first standalone executive comp structure (target multiples, performance vesting, peer group).
Industry Context
External web evidence reinforces three structural facts about the NAND industry that the filings already establish, and adds one piece of color that filings do not.
Reinforced from external sources:
Industry concentration. TrendForce Q3/Q4 calendar 2025 share data confirms the five-player concentration (Samsung, SK Group, Micron, Kioxia, SanDisk) named in SanDisk's 10-K. The Solidigm/SK hynix consolidation is now operational (SK Group at 30%+ enterprise SSD share).
Cycle position. Kioxia FY26 (May 2026) and SK hynix FY25 (January 2026) IR releases both confirm peak-cycle operating margins (37% and 49% respectively). This is not a SanDisk-only optical effect — the entire memory complex is earning supernormal returns.
Capex pivot. Industry-tracker commentary (TrendForce, June 2026) flags a 41% YoY combined capex jump for the SanDisk + Kioxia JV to ~$4.5B. This is the early signal that the supply discipline anchoring 2026 ASPs is starting to bend. Whether peers (Samsung, SK hynix, Micron) follow with proportionally aggressive capex will determine the timing of the next down-leg.
Net new from external sources:
HDD/SSD cost gap. Tom's Hardware reported the datacenter SSD vs. HDD cost-per-TB gap widened from 6x to 16x between Q2 2025 and Q1 2026 — HDD prices up 35% vs. SSDs up 257%. This is meaningful for SanDisk's bit-volume thesis: the cold-tier storage opportunity that SSDs were going to take from HDDs has been pushed out as HDD nearline scarcity defends HDD economics. SanDisk's datacenter mix shift is happening in spite of, not because of, the HDD-substitution narrative.
What the industry external view does not change. The thesis-decisive questions — NBM contract durability, peer capex aggregation, sell-side dispersion on FY27 consensus, HBF schedule — remain the same priority list they were before the web search. The external sources confirmed the up-leg is real and industry-wide; they did not resolve whether the structural reset at the gross-margin floor is durable.
How To Read This Tab
This run's external web research was constrained by a backend availability issue, and that constraint is documented at the top. The findings above are confined to what was retrievable from peer IR filings, SEC EDGAR submissions, and industry-tracker commentary cited in the staged competition-evidence file. Every "External lookup not completed" tag in the specialist Q&A is a deliberate flag that a downstream investor will want resolved before committing capital to a position thesis. The two priority resolutions are (a) NBM contract terms (take-or-pay vs. price-floor mechanics) and (b) Form-4-by-Form-4 insider behavior after the 40x post-spin run-up.